Releases: Azure/AKS-Edge
Releases · Azure/AKS-Edge
1.6.384.0 Release
We are excited to release the February Update of AKS Edge Essentials.
Version Numbers
- AKS-EE Version: 1.6.384.0
- Mariner Version: 2.0.20240123
- K8s (upstream Kubernetes) distribution - 1.26.6, 1.27.6
- K3s distribution – 1.26.10, 1.27.6
- Json version schema - 1.11
New Features
- Kubernetes versions supported: On K8S - 1.26.6 and 1. 27.6, On K3S 1.26.10 and 1.27.6
- Update from K3S 1.26.6 to 1.26.10, and on K8S from 1.26.6 (Nov release) to 1.26.6 (Jan release).
- Upgrade from K3s 1.25.11(Nov release) to 1.26.10(Jan release) and on K8s from 1.25.11(Nov release) to 1.26.6(Jan release)
- We have added json config validation for location field and improved error handling to create a smoother deployment process.
Experimental or Preview Features
- Run Windows containerized workloads.
- Create multi-node deployments either on a single machine or across multiple machines.
- GPU-PV is in preview
For more information feature set and benefits, see AKS Edge Essentials Overview.
Key bug fixes
- Updated journald config to ensure the journal log does not grow beyond the 250-300MB range in order to preserve log space
- Improved install and update experience by optimizing to reduce the free disk space requirement
Known issues
Update
- When updating from 1.5.203 to 1.6.384.0 on both Linux and Windows, customers may see one of the following non-fatal errors:
- "Get-Content: Cannot find path 'C:\Windows\System32\akswindows-manifest.json' because it does not exist."
- "Get-Content: Cannot find path 'C:\Program Files\AksEdge\akswindows-manifest.json' because it does not exist."
This is a non-fatal error and can be ignored.
- Updates from K8S 1.26.3 to 1.26.6 fails when disk pressure is high. As a workaround, extract the tarball /opt/container-images/container-images-tar-file.tar.gz and use ctr to reimport the images that are missing, using the command
ctr -n=k8s.io images import {extracted tar files location}
Deployment
-
New deployment fails when the execution policy is set to AllSigned (
Set-ExecutionPolicy –ExecutionPolicy AllSigned) . Recommend setting the policy to 'ByPass'. -
Uninstalling and reinstalling k8s/k3s causes a “deployment already exists” error or the 'Virtual machine creation failed' error, when creating a new deployment, so recommend restarting your machine before deployment .
-
Event logging fails when uninstalling and reinstalling an MSI without a reboot in between . Particularly true when switching between using K3S and K8S.
-
When a deployment fails due to lack of memory/disk space, it creates an incomplete VHDX file that causes errors in later deployments .
-
Deployment with External virtual switch fails when using Windows RS5.
Linux nodes
- When you use the Remove-AksEdgeNode to remove a node, it might still show up in the
kubectl get nodesas the cluster tries to reconcile the state of the present nodes. Please note that the command Remove-AksEdgeNode is a asynchronous call and will return before the cluster state is updated to reflect the new state.
Windows nodes
- On a K3s Windows node, load balancer klipper-lb pulls images from the internet when connected .
- Adding a Windows node takes around ~10 minutes to complete .
- No storage provisioner is available for the Windows node .
- NFS subdir external provisioner pod fails on Windows nodes.
1.5.203 release
We are excited to release the October Update of AKS Edge Essentials.
Version Numbers
- AKS-EE Version: 1.5.203.0
- Mariner Version: 2.0.20231004
- K8s (upstream Kubernetes) distribution - 1.26.6, 1.25.11
- K3s distribution – 1.26.6, 1.25.11
- Json version schema - 1.9
New Features
- Kubernetes versions supported: On K8S - 1.25.11 and 1. 26.6, On K3S 1.25.11 and 1.26.6
- Update from K3S 1.25.8 to 1.25.11, and on K8S from 1.25.7 to 1.25.11 and from 1.26.3 to 1.26.6
Experimental or Preview Features
- Run Windows containerized workloads.
- Create multi-node deployments either on a single machine or across multiple machines.
For more information feature set and benefits, see AKS Edge Essentials Overview.
Key bug fixes
- Fixed bugs related to network switch impacting deployment experiences.
- Fixed issues with resource group requirements during Arc connectivity.
- Added checks in place, that prevents the windows node update if the machine does not have enough disk space to extract the new windows vhdx.
- Added checks in place to identify scenarios where there are multiple Azure resource types with the same name.
- WssdAgent updated to add subnet information if the switch does not have subnet information.
- Improve deployment experience, by cleaning up artifacts from a previous deployment before starting the new deployment
- Fixed the issue where single machine clusters were ignoring the DNS Server configuration parameter.
Known issues
Update
- Updates from K8S 1.26.3 to 1.26.6 fails when disk pressure is high. As a workaround, extract the tarball /opt/container-images/container-images-tar-file.tar.gz and use ctr to reimport the images that are missing, using the command
ctr -n=k8s.io images import {extracted tar files location}
Deployment
- Deploying a K8s Linux node on a 4GB RAM host machine fails.
- New deployment fails when the execution policy is set to AllSigned (
Set-ExecutionPolicy –ExecutionPolicy AllSigned). Recommend setting the policy to 'ByPass'. - Uninstalling and reinstalling k8s/k3s causes a “deployment already exists” error or the 'Virtual machine creation failed' error, when creating a new deployment, so recommend restarting your machine before deployment.
- Event logging fails when uninstalling and reinstalling an MSI without a reboot in between. Particularly true when switching between using K3S and K8S.
- When a deployment fails due to lack of memory/disk space, it creates an incomplete VHDX file that causes errors in later deployments.
- Deployment with External virtual switch fails when using Windows RS5.
Linux nodes
- When you use the Remove-AksEdgeNode to remove a node, it might still show up in the
kubectl get nodesas the cluster tries to reconcile the state of the present nodes. Please note that the command Remove-AksEdgeNode is a asynchronous call and will return before the cluster state is updated to reflect the new state.
Windows nodes
- On a K3s Windows node, load balancer klipper-lb pulls images from the internet when connected.
- Adding a Windows node takes around ~10 minutes to complete.
- K3s Windows node with Calico as the CNI is not yet supported.
- No storage provisioner is available for the Windows node.
- NFS subdir external provisioner pod fails on Windows nodes.
1.4.109.0 September Release
We are excited to release the September Update of AKS Edge Essentials.
New Features
- Kubernetes 1.26 support. We are now supporting two Kubernetes versions, 1.25 and 1.26.
- Log Partition Size Customizability. Now the log partition size can be set between 1-10GB
Version Numbers
- AKS-EE Version: 1.4.109.0
- Mariner Version: 2.0.20230904
- K8s (upstream Kubernetes) distribution - 1.26.3, 1.25.7
- K3s distribution – 1.26.6+k3s, 1.25.8+k3s
- Json version schema - 1.9
Experimental or Preview Features
- Run Windows containerized workloads
- Create multi-node deployments either on a single machine or across multiple machines.
- Create a cluster with multiple machines
For more information feature set and benefits, see AKS Edge Essentials Overview.
Known issues
Update
- OS upgrade (Windows 10 to Windows 11) will result in AKS Edge Essentials failing to work properly
- "Unable to Start HCS VM" exception caught when running command "Start-AksEdgeControlPlaneUpdate -firstControlPlane True" If this error is encountered, the following commands can be used to recover the cluster
import-module "C:\Program Files\WindowsPowerShell\Modules\1.3.186.0" -Force $nodeConfig = Get-NodeConfig -NodeType ([NodeType]::Linux) $ServiceIpRange = ParseServiceIpRange -ServiceIpRange $nodeConfig.ServiceIpRange $scriptArgs = @{ firstControlPlane = $firstControlPlane networkPlugin = $(Get-ClusterNetworkPlugin) controlPlane = $(Get-LinuxNodeHasControlPlane) controlPlaneEndpointIp = $(Get-ClusterControlPlaneEndpoint).ControlPlaneEndpointIp serviceIPRangeStart = $($ServiceIpRange.Start) serviceIPRangeEnd = $($ServiceIpRange.End) } Update-PostScriptLinuxVM @scriptArgs
Deployment
- Deploying a K8s Linux node on a 4GB RAM host machine fails
- New deployment fails when the execution policy is set to AllSigned (
Set-ExecutionPolicy –ExecutionPolicy AllSigned) - Uninstalling and reinstalling k8s/k3s causes a “deployment already exists” error when creating a new deployment, so a restart is necessary
- Event logging fails when uninstalling and reinstalling an MSI without a reboot in between
- When a deployment fails due to lack of memory/disk space, it creates an incomplete VHDX file that causes errors in later deployments
- Deployment with External virtual switch fails when using Windows RS5
Linux nodes
- When you use the Remove-AksEdgeNode to remove a node, it might still show up in the
kubectl get nodesas the cluster tries to reconcile the state of the present nodes. Please note that the command Remove-AksEdgeNode is a asynchronous call and will return before the cluster state is updated to reflect the new state.
Windows nodes
- On a K3s Windows node, load balancer klipper-lb pulls images from the internet when connected
- Adding a Windows node takes around ~10 minutes to complete
- K3s Windows node with Calico as the CNI is not yet supported
- No storage provisioner is available for the Windows node
- NFS subdir external provisioner pod fails on Windows nodes
- Multi-Node update Fails to recreate Windows nodes
Arc Connection
- Connecting a cluster that has Windows nodes to Arc may fail due to "logcollector*" pod getting stuck in Imageloopbackoff state.
1.3.186.0 August Release
We are excited to release the August Update of AKS Edge Essentials.
New Features
- Kubernetes 1.25 support (1.24 is now deprecated)
- Kubernetes 1.24 to 1.25 upgrade
- Existing deployments on AKS-EE 1.2.414.0 and Kubernetes minor version 1.24 can upgrade to Kubernetes minor version 1.25 by runningSet-AksEdgeUpgrade -acceptUpgrade $TrueSee documentation for more details- Note: Existing deployments with Windows node or Scalable Cluster must FIRST run an additional command in order to upgrade from Kubernetes version 1.24 to 1.25. For these deployments, running Set-AksEdgeUpgrade without setting this reg key will result in failure.
reg add hklm\software\microsoft\aksedge /v acceptUpgrade /t REG_SZ /d True /f
- Note: Existing deployments with Windows node or Scalable Cluster must FIRST run an additional command in order to upgrade from Kubernetes version 1.24 to 1.25. For these deployments, running Set-AksEdgeUpgrade without setting this reg key will result in failure.
Version Numbers
- K8s (upstream Kubernetes) distribution - 1.25.7
- K3s distribution - 1.25.8+k3s
- Json version schema - 1.8
Experimental or Preview Features
- Run Windows containerized workloads
- Create multi-node deployments either on a single machine or across multiple machines.
- Create a cluster with multiple machines
For more information feature set and benefits, see AKS Edge Essentials Overview.
Fixed issues
- Previously, update would fail if there was not enough available disk space. Now, update disk space has been optimized and warning has been added to prevent updates if sufficient disk space is not available.
- Fixed: When disk space begins to run out, precached images may be removed by kubelet/containerd, causing Kubernetes to fail
- Fixed: [BUG] Windows Node status "NotReady" after restarting Host machine #127
- Fixed: IP address '192.168.254.1' is not free, While doing offline installation #131
- Various bug fixes and stability improvements
Known issues
Update
- OS upgrade (Windows 10 to Windows 11) will result in AKS Edge Essentials failing to work properly
- "Unable to Start HCS VM" exception caught when running command "Start-AksEdgeControlPlaneUpdate -firstControlPlane True" If this error is encountered, the following commands can be used to recover the cluster
import-module "C:\Program Files\WindowsPowerShell\Modules\1.3.186.0" -Force $nodeConfig = Get-NodeConfig -NodeType ([NodeType]::Linux) $ServiceIpRange = ParseServiceIpRange -ServiceIpRange $nodeConfig.ServiceIpRange $scriptArgs = @{ firstControlPlane = $firstControlPlane networkPlugin = $(Get-ClusterNetworkPlugin) controlPlane = $(Get-LinuxNodeHasControlPlane) controlPlaneEndpointIp = $(Get-ClusterControlPlaneEndpoint).ControlPlaneEndpointIp serviceIPRangeStart = $($ServiceIpRange.Start) serviceIPRangeEnd = $($ServiceIpRange.End) } Update-PostScriptLinuxVM @scriptArgs
Deployment
- Deploying a K8s Linux node on a 4GB RAM host machine fails
- New deployment fails when the execution policy is set to AllSigned (
Set-ExecutionPolicy –ExecutionPolicy AllSigned) - Uninstalling and reinstalling k8s/k3s causes a “deployment already exists” error when creating a new deployment, so a restart is necessary
- Event logging fails when uninstalling and reinstalling an MSI without a reboot in between
- When a deployment fails due to lack of memory/disk space, it creates an incomplete VHDX file that causes errors in later deployments
- Deployment with External virtual switch fails when using Windows RS5
Linux nodes
- When you use the Remove-AksEdgeNode to remove a node, it might still show up in the
kubectl get nodesas the cluster tries to reconcile the state of the present nodes. Please note that the command Remove-AksEdgeNode is a asynchronous call and will return before the cluster state is updated to reflect the new state.
Windows nodes
- On a K3s Windows node, load balancer klipper-lb pulls images from the internet when connected
- Adding a Windows node takes around ~10 minutes to complete
- K3s Windows node with Calico as the CNI is not yet supported
- No storage provisioner is available for the Windows node
- NFS subdir external provisioner pod fails on Windows nodes
- Multi-Node update Fails to recreate Windows nodes
1.2.414.0 June Release
We are excited to release the June Update of AKS Edge Essentials.
Key Updates
- New version Kubernetes support - K8s (upstream Kubernetes) distribution - 1.25.4 and on K3s distribution - 1.25.8. You cannot update from 1.24.x to 1.25.x using the Microsoft Updates channel, you would have to start afresh.
- Updates on existing version - K8S distribution - 1.24.9, K3S distribution - 1.24.12. You can update your existing 1.24.x deployments to the latest versions on 1.24.-
- On the Windows node, replaced the Windows client image with the ServerCore image. This feature is still in preview.
Experimental or Preview Features
- Run Windows containerized workloads
- Create multi-node deployments either on a single machine or across multiple machines.
- Create a cluster with multiple machines
For more information feature set and benefits, see AKS Edge Essentials Overview.
Fixed issues
- Failure to start WSSD Agent Service when creating a full deployment with auto created external switch - Fixed by waiting for the host adapter to come online, before creating the external switch
- Fixed the issue when you cannot deploy SingleMachineCluster, ARP check does not find available IPs addresses that can be used.
- Fix to ensure that proxy settings works on Windows VM as well
- Fixed the failure of Windows Vhdx extraction on PS 5.0 and 7.0
- Fixed the issue where DNS would not resolve in the Windows node](#100).
Known issues
Runtime
- When disk space begins to run out, precached images may be removed by kubelet/containerd, causing Kubernetes to fail
Update
- OS upgrade (Windows 10 to Windows 11) will result in AKS Edge Essentials failing to work properly
Deployment
- Deploying a K8s Linux node on a 4GB RAM host machine fails
- New deployment fails when the execution policy is set to AllSigned (
Set-ExecutionPolicy –ExecutionPolicy AllSigned) - Uninstalling and reinstalling k8s/k3s causes a “deployment already exists” error when creating a new deployment, so a restart is necessary
- Event logging fails when uninstalling and reinstalling an MSI without a reboot in between
- When a deployment fails due to lack of memory/disk space, it creates an incomplete VHDX file that causes errors in later deployments
- Deployment with External virtual switch fails when using Windows RS5
Linux nodes
- When you use the Remove-AksEdgeNode to remove a node, it might still show up in the
kubectl get nodesas the cluster tries to reconcile the state of the present nodes. Please note that the command Remove-AksEdgeNode is a asynchronous call and will return before the cluster state is updated to reflect the new state.
Windows nodes
- On a K3s Windows node, load balancer klipper-lb pulls images from the internet when connected
- Adding a Windows node takes around ~10 minutes to complete
- K3s Windows node with Calico as the CNI is not yet supported
- No storage provisioner is available for the Windows node
- NFS subdir external provisioner pod fails on Windows nodes
- Multi-Node update Fails to recreate Windows nodes
1.1.220.0 April Release
We are excited to release the April Update of AKS Edge Essentials. This release adds support for volume licensing based model, which offers a more traditional licensing model for customers who need to deploy AKS Edge Essentials in disconnected scenarios and can't report their usage to Microsoft. An annual subscription can be purchased for all the devices in your AKS Edge Essentials cluster, and allows unlimited use of AKS Edge Essentials, on a specified number of devices, without the need to report usage to Azure commerce. For more information visit http://aka.ms/aks-edge/licensing
Key features
- Existing features from previous (GA) release
- K8s (upstream Kubernetes) distribution - 1.24.6
- K3s distribution - 1.24.6
- Volume licensing based model support
Experimental or Preview Features
- Run Windows containerized workloads
- Create a dual-node Linux node (control plane and worker) + Windows node (worker) cluster on a single machine
- Create a cluster with multiple machines
For more information feature set and benefits, see AKS Edge Essentials Overview.
Fixed issues
- Non-english language support
- Miscellaneous bug fixes
Known issues
Runtime
- When disk space begins to run out, precached images may be removed by kubelet/containerd, causing Kubernetes to fail
Update
- OS upgrade (Windows 10 to Windows 11) will result in AKS Edge Essentials failing to work properly
Deployment
- Deploying a K8s Linux node on a 4GB RAM host machine fails
- New deployment fails when the execution policy is set to AllSigned (
Set-ExecutionPolicy –ExecutionPolicy AllSigned) - Uninstalling and reinstalling k8s/k3s causes a “deployment already exists” error when creating a new deployment, so a restart is necessary
- Event logging fails when uninstalling and reinstalling an MSI without a reboot in between
- When a deployment fails due to lack of memory/disk space, it creates an incomplete VHDX file that causes errors in later deployments
- Deployment with External virtual switch fails when using Windows RS5
Windows nodes
- On a K3s Windows node, load balancer klipper-lb pulls images from the internet when connected
- Adding a Windows node takes around ~10 minutes to complete
- K3s Windows node with Calico as the CNI is not yet supported
- No storage provisioner is available for the Windows node
- NFS subdir external provisioner pod fails on Windows nodes
- Multi-Node update Fails to recreate Windows nodes
1.1.80.0 March Update
We are excited to release the March Update of AKS Edge Essentials, a lightweight Kubernetes platform that enables rapid application innovation and modernization at scale at the edge. This release adds new features including TPM Passthrough, dual-NIC support, and non-root access for logs, improved Arc connectivity, and updated Kubernetes version.
Key features
- Existing features from previous (GA) release
- K8s (upstream Kubernetes) distribution - 1.24.6
- K3s distribution - 1.24.6
- TPM Passthrough
- Dual-NIC support
- Improved Arc Connectivity (Az.ConnectedKubernetes v0.9.0)
Experimental or Pre-release Features
- Run Windows containerized workloads
- Create a dual-node Linux node (control plane and worker) + Windows node (worker) cluster on a single machine
- Create a cluster with multiple machines
For more information feature set and benefits, see AKS Edge Essentials Overview.
Fixed issues
- No longer requires root user/group to access container log files
- Miscellaneous bug fixes to improve stability
Known issues
Runtime
- When disk space begins to run out, precached images may be removed by kubelet/containerd, causing Kubernetes to fail
Update
- OS upgrade (Windows 10 to Windows 11) will result in AKS Edge Essentials failing to work properly
Deployment
- Deploying a K8s Linux node on a 4GB RAM host machine fails
- New deployment fails when the execution policy is set to AllSigned (
Set-ExecutionPolicy –ExecutionPolicy AllSigned) - Uninstalling and reinstalling k8s/k3s causes a “deployment already exists” error when creating a new deployment, so a restart is necessary
- Event logging fails when uninstalling and reinstalling an MSI without a reboot in between
- When a deployment fails due to lack of memory/disk space, it creates an incomplete VHDX file that causes errors in later deployments
- Deployment with External virtual switch fails when using Windows RS5
Windows nodes
- On a K3s Windows node, load balancer klipper-lb pulls images from the internet when connected
- Adding a Windows node takes around ~10 minutes to complete
- K3s Windows node with Calico as the CNI is not yet supported
- No storage provisioner is available for the Windows node
- NFS subdir external provisioner pod fails on Windows nodes
- Multi-Node update Fails to recreate Windows nodes
1.0.406.0 Generally Available (GA)
We are excited to announce the GA of AKS Edge Essentials, a lightweight Kubernetes platform that enables rapid application innovation and modernization at scale at the edge. In addition, you can connect your AKS Edge Essentials cluster to Azure via Arc-enabled Kubernetes, enabling robust deployment, management, and governance capabilities for your IoT devices using the Azure portal.
Key features
- K8s (upstream Kubernetes) distribution - 1.24.3
- K3s distribution - 1.24.3
- Run Linux containerized workloads
- Create a single-node Linux (control plane and worker) cluster on a single machine:
-deploymenttype : singlemachineclusterfor Internal virtual swtich
-deploymenttype : scalableClusterfor External virtual switch - Remotely deploy your workloads using Azure Arc GitOps
- Remotely deploy Azure services using Azure Arc
- Use calico (K8s) or flannel (K3s) network plugins
- Installation without internet connectivity and proxy enabled environments
- Updates available through Microsoft Update and Microsoft Update Catalog
Experimental or Pre-release Features
- Run Windows containerized workloads
- Create a dual-node Linux node (control plane and worker) + Windows node (worker) cluster on a single machine
- Create a cluster with multiple machines
For more information feature set and benefits, see AKS Edge Essentials Overview.
Fixed issues
- #20 - AKS Arc agent connection fails with timeout (aad-proxy)
- #65 - Quickstart script failed to connect to Azure Arc
Known issues
Runtime
- When disk space begins to run out, precached images may be removed by kubelet/containerd, causing Kubernetes to fail
Update
- OS upgrade (Windows 10 to Windows 11) will result in AKS Edge Essentials failing to work properly
Deployment
- Deploying a K8s Linux node on a 4GB RAM host machine fails
- New deployment fails when the execution policy is set to AllSigned (
Set-ExecutionPolicy –ExecutionPolicy AllSigned) - Uninstalling and reinstalling k8s/k3s causes a “deployment already exists” error when creating a new deployment, so a restart is necessary
- Event logging fails when uninstalling and reinstalling an MSI without a reboot in between
- When a deployment fails due to lack of memory/disk space, it creates an incomplete VHDX file that causes errors in later deployments
- Deployment with External virtual switch fails when using Windows RS5
Windows nodes
- On a K3s Windows node, load balancer klipper-lb pulls images from the internet when connected
- Adding a Windows node takes around ~10 minutes to complete
- K3s Windows node with Calico as the CNI is not yet supported
- No storage provisioner is available for the Windows node
- NFS subdir external provisioner pod fails on Windows nodes
- Multi-Node update Fails to recreate Windows nodes
1.0.358.0 (Public Preview) Update MSI Release Only
This release is the third AKS-EE public preview release. Please note that this release is Update MSI release only, meaning customers must update from1.0.266.0 to get to 1.0.358.0. There is no mechanism to install 1.0.358.0 outright. Please also note that the changes in this release are relatively small, and largely intended to give customers some experience with the update process.
Key features
- Update available through Microsoft Update and Microsoft Update Catalog. For more information on updates please visit AKS-EE Updates.
- Windows nodes continue to be an experimental feature
- Updated warnings and log outputs
- K8s (upstream Kubernetes) distribution - 1.24.3
- K3s distribution - 1.24.3
Fixed issues
- Fixed: Default proxy settings in the aksedge-dc.json template file is preventing computed no_proxy configuration from working
Known issues
Runtime
- When disk space begins to run out, precached images may be removed by kubelet/containerd, causing Kubernetes to fail
Update
- OS upgrade (Windows 10 to Windows 11) will result in AKS Edge Essentials failing to work properly
Deployment
- Deploying a K8s Linux node on a 4GB RAM host machine fails
- New deployment fails when the execution policy is set to AllSigned (
Set-ExecutionPolicy –ExecutionPolicy AllSigned) - Uninstalling and reinstalling k8s/k3s causes a “deployment already exists” error when creating a new deployment, so a restart is necessary
- Event logging fails when uninstalling and reinstalling an MSI without a reboot in between
- When a deployment fails due to lack of memory/disk space, it creates an incomplete VHDX file that causes errors in later deployments
Windows nodes
- On a K3s Windows node, load balancer klipper-lb pulls images from the internet when connected
- Adding a Windows node takes around ~10 minutes to complete
- K3s Windows node with Calico as the CNI is not yet supported
- No storage provisioner is available for the Windows node
- NFS subdir external provisioner pod fails on Windows nodes
- GitOps steps provided in the documentation are not compatible with Windows nodes
1.0.266.0 (Public Preview)
This version is a public preview refresh and a release candidate for our final GA bits.
Key features
- Improved stability of both Linux and Windows nodes
- Windows nodes continue to be an experimental feature
- Refactored AKSEdge Powershell module with newly exposed cmdlets and new JSON configuration files. The new JSON parameters can be found here.
- Support for external NFS storage provisioned
- K8s (upstream Kubernetes) distribution - 1.24.3
- K3s distribution - 1.24.3
Fixed issues
- #15 - Hyper-V Internal Switch is lost after a restart on Win11 Pro 22H2 22623.746
- #34 - K3s New-AksEdgeDeployment fails on current PowerShell versions (7.3.x)
- #40 - Allow container images to be loaded from archives
- #43 - Calico strictAffinity is false when installing Windows Node
- Memory leak on K3s cluster impacts long duration run of K3s Linux nodes
Known issues
Runtime
- When disk space begins to run out, precached images may be removed by kubelet/containerd, causing Kubernetes to fail
Update
- OS upgrade (Windows 10 to Windows 11) will result in AKS Edge Essentials failing to work properly
Deployment
- Deploying a K8s Linux node on a 4GB RAM host machine fails
- New deployment fails when the execution policy is set to AllSigned (
Set-ExecutionPolicy –ExecutionPolicy AllSigned) - Uninstalling and reinstalling k8s/k3s causes a “deployment already exists” error when creating a new deployment, so a restart is necessary
- Event logging fails when uninstalling and reinstalling an MSI without a reboot in between
- When a deployment fails due to lack of memory/disk space, it creates an incomplete VHDX file that causes errors in later deployments
Windows nodes
- On a K3s Windows node, load balancer klipper-lb pulls images from the internet when connected
- Adding a Windows node takes around ~10 minutes to complete
- K3s Windows node with Calico as the CNI is not yet supported
- No storage provisioner is available for the Windows node
- NFS subdir external provisioner pod fails on Windows nodes
- GitOps steps provided in the documentation are not compatible with Windows nodes